Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Changes in benthic marine macrophyte community structure in the Strait of Georgia: long-term and grazing… Manson, Murray 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_manson_murray.pdf [ 2.99MB ]
JSON: 831-1.0086259.json
JSON-LD: 831-1.0086259-ld.json
RDF/XML (Pretty): 831-1.0086259-rdf.xml
RDF/JSON: 831-1.0086259-rdf.json
Turtle: 831-1.0086259-turtle.txt
N-Triples: 831-1.0086259-rdf-ntriples.txt
Original Record: 831-1.0086259-source.json
Full Text

Full Text

CHANGES IN BENTHIC MARINE MACROPHYTE COMMUNITYSTRUCTURE IN THE STRAIT OF GEORGIA:LONG-TERM AND GRAZING RESPONSESbyMURRAY M. MANSONB.Sc., The University of Western Ontario, 1989A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Department of Botany)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAOctober 1993© Murray M. Manson, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of 80 4ct, A-1 LiThe University of British ColumbiaVancouver, CanadaDate Dc-±, lq ) 113DE-6 (2/88)ABSTRACTBenthic marine macrophyte community structure at twosites near Bath and Sear Islands in the Strait of Georgia,British Columbia, was quantified from the high tide level onthe shore to about 10 m below zero tide level. Data obtainedfrom the sites during the summers of 1972 through 1975 and in1992 was used to make several comparisons of macrophytecommunity structure. 1) Changes in community structurealong the depth gradient from intertidal to subtidal wereinvestigated. 2) At Bath Island, a period of intensivegrazing by sea urchins, Strongylocentrotus droebachiensis,during 1973 altered the community structure and provided anopportunity for a comparative study of macrophyte communitychange and regeneration at the two sites. 3) Communitystructure at both sites in 1992 was compared to that in the1970's. A multivariate approach was taken for the analyses,using principal component analysis, canonical variateanalysis (CVA), and a modification of CVA including depth asan environmental variable. Several macrophyte associationswere consistently found by the analyses.1) An association of Fucus spp., Mastocarpus papillatus,Neorhodomela larix, Porphyra spp., and Ulva spp. defined thecommunity structure of intertidal quadrats. 2) A subtidalfoliose red and kelp species association was found inungrazed areas. 3) A grazed form of the subtidalassociation was characterized by low variability and highiiabundances of Calliarthron tuberculosum and Ulvaria obscure.4) After grazing pressure was removed, community structurewas characterized by initial increases in Ulva spp. andSargassum muticum during the first year, followed by regrowthof foliose red and kelp species. 5) A 1992 upper subtidalassociation characterized by high abundances of S. muticumand Lomentaria hakodatensis, Gigartina exasperate, Corallinaofficinalis, and Gelidium spp. contrasted with a 1970'sassociation of such species as Callophyllis spp.,Cryptopleura ruprechtiana, Polyneura latissima, andDesmarestia viridis.TABLE OF CONTENTSPageABSTRACT^ iiTABLE OF CONTENTS^ ivLIST OF TABLESLIST OF FIGURES^ viACKNOWLEDGEMENTS viiiINTRODUCTION^ 1MATERIALS AND METHODS^ 9Study Area^ 9Sampling Procedure 9Statistical Analysis^ 12Principal Components Analysis^ 14Canonical Variate Analysis 16Canonical Correlation Analysis 18RESULTS^ 19Distributional Properties of^ 19the DataPrincipal Components Analysis 22Canonical Variate Analysis^ 33Canonical Correlation Analysis 42DISCUSSIONCommunity Change Along theDepth GradientCommunity Change and SeaUrchin Grazing ActivityCommunity Change After SeaUrchin RemovalCommunity Change from the1970's to 1992CONCLUSIONSBIBLIOGRAPHYAPPENDIX 15555575961697175ivLIST OF TABLESTable1. Sampling dates and number of quadrats sampledat each site.2. Eigenvalues, percent variance accounted for,and component loadings for the first threeaxes from the principal components analysisof raw species abundances in all quadrats.Page10233. PCA of log-transformed species data.^ 254^Statistics describing the 95% confidence^32ellipses from PCA of log-transformed data.5^Canonical correlation coefficients,^ 34redundancies, % variance extracted, andcanonical loadings from the CVA of log-transformed species abundances.6 Canonical correlation coefficients,^ 43redundancies, % variance extracted, andcanonical loadings from the CCO of log-transformed species abundances.VLIST OF FIGURESFigure^ Page1. Number and proportion of sea urchin test^20diameters (cm) collected from Bath Island in1992.2. Q-Q probability plots of a) raw species^21abundances and b) log-transformed speciesabundances.3. Quadrat scores on the first principal axis^27from the PCA of log-transformed speciesabundances vs. quadrat depth (m).4. Quadrat scores on the second principal axis^28from the PCA of log-transformed speciesabundances vs. quadrat depth (m).5. Quadrat scores on second principal axis vs.^30scores on first principal axis from the PCAof log-transformed species abundances.6. Quadrat scores on vi and u i axes from the CVA^39of log-transformed species abundances.7. Scatter plots of quadrat scores on u l , u2 and u3^41from the CVA of log-transformed species data.8. Quadrat scores on vi from the canonical^45correlations analysis of log-transformedspecies abundances, site/year dummy variablesand depth vs. quadrat depth (m).9. Quadrat scores on u i from the canonical^50correlations analysis of log-transformedspecies abundances, site/year dummy variablesand depth vs. quadrat depth (m).10. Quadrat scores on u2 and v2 from the canonical^52correlations analysis of the log-transformedspecies abundances, site/year dummy variablesand depth.11. Quadrat scores on u 3 and v3 from the canonical^54correlations analysis of the log-transformedspecies abundances, site/year dummy variablesand depth.vi12. Time-series of seawater surface temperatures^66at Active Pass, British Columbia, near thestudy sites.viiACKNOWLEDGEMENTSI would like to thank my supervisor, Dr. Ron Foreman,for making this work possible. He supplied the samplingequipment, field expenses, and for the most part, the data.Acknowledgements must be extended to the many people whoassisted with the nearshore benthic ecosystems project duringthe 1970's. I am also grateful for the financial assistanceof Environment Canada, Fisheries and Marine Service Grant-in-Aid of University Research and National Research Council ofCanada Operating Grant No. A6241, which were the sources ofDr. Foreman's funding during the 1970 1 s.The data could never have been collected without thetremendous efforts of Mr. Gavin K. Manson, Mr. JustinO'Connor, and Mr. Matthew Witt, who volunteered to scrapeseaweed off slippery rocks in leaky dry-suits. Your G.I.Joe-like heroics are truly admirable.I would also like to thank Dr. Rob De Wreede, Dr. GaryBradfield, and Dr. Sandra Lindstrom for input andencouragement, and for critically reviewing the drafts ofthis manuscript. Thanks must also be given to Mrs. JulieCelestino-Olivera of the U.B.C. Phycological Herbarium whowas very patient, and provided a great deal of assistance inspecies identification. The time-series temperature data forActive Pass was generously provided by Howard Freeland atIOS.Also of equal importance was Debbie Muggli, whosededication as a climbing and ski partner allowed me to workwithout distraction between trips.viiiINTRODUCTIONThe Strait of Georgia, located between the easternshore of Vancouver Island and the coast of mainland BritishColumbia, is in immediate proximity with the majority of thepopulation of B.C. and Washington State. Heavy use by bothrecreational and commercial vessels, fisheries, and effluentdisposal facilities results in the inevitable occurrence ofdisturbances to the biota of the strait. A further source ofpotential disturbance comes from communities located alongthe Fraser River and its tributaries. Freshwater runoff fromthe Fraser River has a strong influence on the southern halfof the Strait of Georgia, so the quality of this runoff alsohas the potential to impact the biota of the strait. Becauseof this heavy traffic and the significant importance of theStrait of Georgia to the economy of B.C., techniques forstudying disturbances to the natural biota should bedeveloped. Without the ability to assess the extent ofchange caused by a particular disturbance event, there is nopossibility to determine the success of cleanup efforts orthe cost of a disturbance to the various industries withfinancial interest in the biota of the strait.The concept of disturbance and recovery has beeninvestigated experimentally by many marine ecologists,particularly in benthic marine communities (Dethier 1984,Sousa 1985, Witman 1985, deEston and Bussab 1990). Acondition that removes much of the living biomass from a1community (as disturbance will be defined here) can beimposed experimentally, and the effects on communitystructure followed for as long as necessary. The communitystructure may be altered in several ways after such adisturbance (Connell 1987), 1) all species removed by thedisturbance are replaced by new species; 2) gradualreplacement of colonizing species between disturbances eithera) in a predictable, progressive succession of species; or b)in an unpredictable, non-successional pattern of species.The essence of successional change has been the subjectof intensive debate among ecologists, with little clearagreement. Marine ecological field experiments in particularhave not supported the traditional models of succession(Sousa 1980). For example unpredictable episodic events seemto be important to succession both in the creation of space(Dethier 1984, Sousa 1984) and in the events controllingdispersal of propagules (Reed et al. 1988, Reed et al. 1992),making succession unpredictable. The availability ofpropagules at the time of disturbance is another stochasticfactor that can lend to the unpredictable nature of marinebenthic successions (Lubchenco and Menge 1978). Also, earlysuccessional species may have varying effects on thecolonization of later species so that a single model ofsuccession is not tenable (Farrell 1991).One particular aspect of disturbance in marine hard-substrata that has received considerable attention frommarine ecologists is that of the effects of sea urchin2grazing on kelp community structure. Paine and Vadas (1969)describe how intensive grazing by sea urchins reduces thediversity of algal assemblages during grazing, but increasesthe diversity during recolonization. This work seems to havespurred the development of the widely accepted sea otterkeystone species hypothesis in the Northeast Pacific (Estesand Palmisano 1974, Estes et al. 1978), and possibly theconcurrent lobster keystone species hypothesis of theNorthwest Atlantic discussed by Elner and Vadas (1990). Amyriad of work has generally concluded that intensive grazingby sea urchins can create "urchin barrens" (i.e. Pearse andHines 1979, Duggins 1981, Harrold and Reed 1985, Dean et al.1989), and removal of the urchins results in thereestablishment of kelp dominated communities (Foreman 1977).The process of disturbance has been distinguished fromthat of stress, which does not result in the active removalof biomass (thus creating new space), but limits the growthor reproduction of a selection of species in the community(Clayton 1990). Both disturbance and stress could be presentin the natural environment, affecting community structure,but different approaches are used in experimentalinvestigations of these two conditions (Bender et al. 1984).The investigation of stress conditions is further complicatedby the relative nature of stress (one species may be stressedwhile others are unaffected or benefit), the variationpossible in the intensity of stress, and the difficulty indiscerning the compounding effects that stress conditions may3create. Gradually changing environmental conditions aredifficult to measure, and their effects on speciescomposition may be noticeable only over periods of severaldecades. The long-term studies needed to investigate theseeffects are costly and rarely carried out but can showchanges in community that would otherwise go unnoticed (i.e.changes in marine intertidal communities along the EnglishChannel, Southward 1976).To investigate a disturbance event or the introductionof a stress condition in the natural environment, the luxuryof preplanning is rarely possible, and the process ofinvestigating the effects of the condition becomesconsiderably more difficult. In order to accurately assessthe environmental impact caused by a natural disturbanceevent, some important information must be included in thestudy. There must be a quantitative assessment of the biotaat the disturbed site before the disturbance event, to act asa baseline to compare the post-disturbance character of thesite (Green 1979). Simple presence/absence data can fail toshow change, as the relative abundances of species may be allthat changes. Information of this sort is difficult andcostly to obtain, especially over large areas, and it mayrequire a great deal of luck (or persistant sampling) to havethis baseline data for the right place at the time of thedisturbance. Also essential to the study is a site ofsimilar quantitative character that escaped the disturbance.The data from this site are used as a control, to indicate4the degree of change which is occurring at the disturbedsite, unrelated to the disturbance (Green 1979).The ease of obtaining quantitative information of aparticular community ultimately depends on the habits of theorganisms studied, and the limits chosen for the number ofspecies to be included in the study. Macrophytes inparticular provide sessile, complex communities that can berelatively easily sampled in a random design, and used toindicate the effects of a disturbance. The difficulty insampling increases with subtidal sampling, relatively fewquantitative studies having been done that include subtidalsampling of the entire depth gradient over which macrophytescan occur (Foster 1990, Chapman 1986).An important aspect of an environmental impactassessment involves the techniques used to define thecommunity structure at the study sites. The data obtained instudies of community structure are multivariate and highlycomplex, often not meeting distributional assumptionsrequired for some statistical techniques (Green 1979, Jongmanet al. 1987), making interpretation difficult. Thetechniques used for interpretation of the data are thusimportant to the information gained from the data. Asmentioned earlier, in order to assess the extent of change incommunity structure, the data must be quantitative, so theanalysis must consider the quantitative nature of the data.Further, important information on interactive effects can beoverlooked by assessing a multivariate data set as a series5of univariate responses. Multivariate methods of analysisare thus advantageous in showing community responses tovarying conditions. Currently the most common techniquesused by environmental biologists for quantitativemultivariate data involve either cluster analysis or someform of ordination (Green 1979, Jongman et al. 1987).Relatively little work has been done involvingcomparative multivariate analyses of marine macrophytecommunities. Cluster analysis was used to compare macrophytecommunities after experimental manipulation (deEston andBussab 1990), and to investigate the effects of sewageeffluent on species assemblages (Littler and Murray 1975).Cluster analysis was also one of the techniques used byForeman (1977), in his investigation of macrophyte communityregeneration following grazing by sea urchins. Clusteranalysis provides an interesting summary of similarity amongquadrats, but does not indicate the basis for similarity.Ordination techniques have the advantage of providingthe loadings that each species has on the quadrat scores.Correspondence analysis (Hill 1973) was used to successfullydistinguish between subtidal quadrats, intertidal quadrats,and quadrats located on different substrata in the ordinationof west African macrophyte vegetation (John et al. 1977).Macrophyte communities on different substrata were assessedon chalk cliffs in England (Tittley and Shaw 1980) usingcorrespondence analysis and another ordination technique,principal coordinate analysis (Pimental 1979).6Multidimensional scaling has also been used to show aseparation between community structure of inshore andoffshore kelp communities (Velimirov et al. 1977). The useof multivariate statistics in comparative studies has beenused successfully in studies of terrestrial plant communities(Gibson and Brown 1992, Gittins 1985), and aquatic benthicinvertebrate studies (Salmon and Green 1983).In this study, I have the opportunity to investigatethe change in macrophyte community structure following anatural disturbance event. The need for a control site andpre-disturbance quantitative data has been met as a long-termquantitative study of benthic community structure andfunction was underway at two similar sites in the Strait ofGeorgia at the time of the disturbance (Foreman 1977,Lindstrom and Foreman 1978). The disturbance event was aperiod of intensive grazing by the green sea urchin,Strongylocentrotus droebachiensis. As the data documentingthis disturbance were collected from 1972 through 1975, theopportunity to compare the current (1992) macrophytecommunity structure at the same sites presents itself. Afurther aspect of this study will thus be to identify changesin community structure that have occured at the study sitesfrom the 1970's to 1992, indicating how appropriate theexisting baseline data would be in future assesments ofdisturbance events at the sites.In carrying out the two aspects of this study, thefollowing objectives will be met. 1) Macrophyte species7associated with a particular location on an intertidal tosubtidal depth gradient will be identified. 2) To identifythe direct effects of sea urchin grazing activity onmacrophyte community structure, the macrophyte speciescomposition and abundance will be compared at Bath and SearIslands before (1972) and during (1973) the urchin grazingactivity. 3) Community regeneration will also beinvestigated by comparing macrophyte community structures ofBath and Sear Island in 1972 to those in 1974 and 1975. 4)The final objective of the study will be to identify long-term changes by comparing the macrophyte community structurein 1972-1975 and in 1992. Throughout the study the focuswill be on the community response, rather than that ofindividual species, by using a multivariate approach to theanalysis.8MATERIALS AND METHODSStudy areaThis research consists of a conjunction of samplingefforts carried out in the summers of 1972 through 1975, andin the summer of 1992. All of these sampling sessionsutilized the same two one-hectare study sites located on thesouth east shores of Bath Island and Sear Island in the FlatTop Islands area (49 ° 09' N x 123 ° 40' W) of the Strait ofGeorgia. The two study sites are located within 1 km of eachother and share the same aspect, though the Bath Island sitemay be slightly more exposed to open water in the Strait ofGeorgia. At both sites, the substratum consists of solidsandstone bedrock that is covered with boulders at greaterdepths. At Bath Island, these boulders are found at about4.5 m below zero tide, some 70 m out from the high water lineon the shore. The Sear Island site is of slightly steeperincline, merging into boulders at 6 m below zero tide line,70 m out from the high water line on shore.Sampling ProceduresDetails of the 1972 through 1975 sampling sessions aredescribed by Foreman (1977), the sampling dates and number ofquadrats sampled during these sessions summarized in Table 1.The following is a description of the sampling protocol9Table 1. Sampling dates and number of quadratssampled at each site.Site^Year^Sampling #QuadratsDatesBath Island^1972^5/18-7/4^391973^5/15-7/26 411974^4/8-7/31^331975 7/7-7/9 191992^6/22-7/4^63Sear Island^1972^5/31-7/31^4019731974^4/15-8/12^321975^7/9-7/10 201992^6/25-7/16^63employed during the 1992 sampling session, designed tofacilitate comparison with the data collected from 1972through 1975. A total of six 100 m transects (three persite) were sampled in the summer of 1992, between June 22 andJuly 3. Each transect ran perpendicular to the shore from arandomly determined point located along the high water line.Quadrats (one-quarter m2 ) were placed at 5 m intervals alongthe transects where the following variables were recorded; 1)the substratum type was assessed as one of four classes(class 1-sand, class 2-shells, class 3-boulders, class 4-continuous solid bedrock), 2) the depth was recorded (latertransformed to meters below zero tide level), 3) the numberof sea urchins was recorded, 4) all macrophytes were removedfrom the substratum via a scraper and an air-driven vacuum asdesigned by Foreman (1977). A total of 126 samples werecollected in 1992 (63 samples from both Bath Island and SearIsland). Following collection, the samples were preserved in5 percent formalin in seawater until further processing couldbe carried out. Processing entailed sorting each sample intoits constituent species and recording the wet weight per m2for each. All samples were processed within 5 months ofcollection. Voucher specimens were prepared for the U.B.C.Phycological Herbarium, and are available there forinspection.For the purposes of the Foreman (1977) study of benthiccommunity recovery following urchin grazing, urchins wereremoved from the Bath Island study site in the summer of 197311by divers. The effort was continued until after sampling wasfinished in 1975, to remove the urchins that had moved intothe area from outside the perimeter of the study site. Toestimate the size distribution of the sea urchin(Strongylocentrotus droebachiensis) population at Bath Islandin 1992, all sea urchins within 1 m on either side of thesecond 100 m transect were collected. Test diameters(millimeters) and weight (grams) were recorded for eachurchin. The urchins were then returned to the site ofcollection.Statistical AnalysesAn important assumption underlying the application ofmany statistical analyses is that the variables be jointlylinear, and have a normal (or multinormal) distribution(Green 1979). To determine if the species biomass datacollected during this study could validly be tested with thestatistical techniques described below, the distributionalproperties of the data were assessed using the graphicaltechnique of Q-Q probability plots (Gnanadesikan 1977). Tocreate these plots for a multivariate data set, the data mustfirst be summarized in a scalar form. The species'abundances in each quadrat were described as the Mahalanobisdistance between each quadrat and the group centroid. Thisunivariate summary was plotted against a chi-squareddistribution with 26 degrees of freedom. A comparison was12made between Q-Q probability plots for both the raw data andlog-transformed (natural log) data to determine which wouldmore closely approach linearity. Based on the success oflog-transformation to improve the joint linearity of thevariables, log-transformed species biomass values were usedfor the following analyses. Multinormality was rejected forboth the raw and log-transformed data sets, so formal testsof dimensionality and equality of vector means were notconsidered in the analysis (Gittins 1985). The techniquesused are considered to be rigorous to violations ofmultinormality, the most important assumption being the jointlinearity of variables (Gittins 1985).Due to the large number of species encountered, thedata were well-suited to analysis using multivariatetechniques. Multivariate analyses provide insight into jointspecies responses, therefore emphasizing community levelinterpretation of the data. As the purpose of this study wasto investigate macrophyte community structure, a multivariateapproach was chosen for the analysis. Some of the featuresof principal components analysis (PCA), canonical variateanalysis (CVA), and canonical correlation analysis (CCO) wereparticularly appropriate for interpretation of this data setin terms of the objectives of the study.13Principal Component AnalysisThe ability of PCA to order quadrats along a singleaxis based on the relative abundance of several importantspecies results in a reduction in dimensionality of the dataset. In this manner, quadrats with similar locations on theaxis have similar species compositions (Jongman et al. 1987).By using the quadrat scores on the component axes asvariables, interpretation of complex data sets is simplified,and joint species responses are emphasized.The component loadings from a PCA show how each speciesinfluences the axes, so species that have similar loadingscan be interpreted as associations (Jongman et al. 1987).The loadings were used to define the species associations inthis study.The focus of this study on comparing the macrophytecommunity structure at Bath and Sear Islands during differentyears was addressed by summarizing the component scores ofthe quadrats from each site/year group (i.e. Bath Island/1972(B72), Bath Island/1992 (B92), Sear Island/1972 (S72)), andusing descriptive statistics and graphics to facilitate theircomparison. These statistics are based on 95 percentconfidence intervals of the component scores for eachsite/year (Jolicoeur and Mosimann 1960). The confidenceintervals can be calculated for each site/year on eachcomponent axis, and graphically represented as ellipses. Theellipses for each site/year can then be compared in terms of14their location on the axes, area (A), eccentricity (ecc), andangle (0) of intersection of the major axis of the ellipsewith the first component axis. The area of the ellipseprovides an indication of the amount of variation in thequadrat scores within the site/year. Eccentricity is ameasure of the shape of the ellipse in terms of the relativelengths of the major and minor axes (ecc = a-1( a2-b2)1/2 ;where 'a' is half the length of the major axis, and 'b' halfthe length of the minor axis), and is bounded by 0 and 1.Biologically, eccentricity represents the extent to which thescores within the site/year on the component axes arecorrelated to one another; eccentricity values approaching 1indicating that the presence or absence of one speciesassociation is strongly correlated with that of another. Theangle 0 indicates the relative extent to which each componentinfluences the spread of points within the site/year.Similar values for 0 at two site/years would indicate similarspecies compositions at those site/years.PCA was carried out on the correlation matrices of twodifferent subsets of the data to show the major associationsof macrophytes; 1) all quadrats, all species encountered morethan 15 times in the data set, 2) quadrats between 3 m aboveand 6 m below zero tide, species encountered more than 15times (excluding most microscopic filamentous andpolysiphonous species) with high component loadings on one ofthe first three component axes from the previous analysis.By using the correlation matrices (Pearson correlation15coefficients), the size differences among macrophyte specieswould not influence the associations (Pielou 1984). Althoughthis method may disproportionately weight the importance ofrare species, it was necessary because the differences inbiomass among macrophytes were so great that the first threeaxes derived from a covariance matrix described variation inonly the large kelp species. Further, by minimizing thenumber of rare species in the analysis, this effect should bereduced.Canonical Variate AnalysisCanonical ordination techniques have the advantage ofperforming an ordination on two sets of related data, suchthat the correlation between the two ordinations is maximized(Gittins 1985). One common use of canonical ordination amongecologists is canonical correlation analysis (CCO) whichinvolves correlating a set of species abundances with a setof environmental variables measured for the same quadrats.Thus, CCO shows the patterns of species variation that moststrongly correlate to patterns of variation in theenvironmental data.Canonical variate analysis (CVA) is a variation of CCOwhere dummy variables are assigned to represent specificgroups of quadrats. The analysis is then carried out as aCCO, with the dummy variable set substituted for theenvironmental data set. Patterns of variation in the species1.6abundance data are thus correlated to the quadrat groups tomaximize the ratio of the between group sum of squares ofquadrat scores to the within group sum of squares along eachaxis (Jongman et al. 1987). The CVA thus emphasizes thedifferences in scores between the groups of quadrats.CVA was useful to this study since part of theobjective was to compare the species wet-weights recorded forquadrats grouped according to the site/year from which theywere collected. The axes resulting from this ordination,calculated to maximize the correlation between species wet-weights and site/year quadrat groups, should emphasize thedifferences in macrophyte species composition and abundanceamong the site/years. When viewed in light of the grazinghistories of the site/years, information on how grazingaffects macrophyte community structure can be inferred.CVA was performed to correlate the same log-transformedspecies data set used in the second PCA with an 8 variablebinary code used to delineate the site/year groups ofquadrats. For example, quadrats sampled at Bath Islandduring 1973 were coded with a score of 1 for the first dummyvariable and 0 for the other seven (i.e. 1,0,0,0,0,0,0,0),while B74 was coded with 0 for the first variable, 1 for thesecond, and 0 for the others (i.e. 0,1,0,0,0,0,0). The ninthsite/year (B72) received 0 for all dummy variables (i.e.0,0,0,0,0,0,0,0).17Canonical Correlation AnalysisThe above CVA was also carried out in a modified form,with depth included as a ninth variable in addition to thedummy variable set. Otherwise, this analysis used the samedata as the CVA. This was done to show how speciescomposition compared among the site/years along the depthgradient.18RESULTSA list of all of the species encountered in the studyis included in Appendix 1. A total of 158 species wereencountered during the 5 years of sampling. Figure 1 showsthe size distribution of Strongylocentrotus droebachiensisat Bath Island in 1992.Distributional Properties of the DataThe Q-Q probability plots are shown in Figure 2. Thelog-transformed species abundances (wet weights) show asubstantial improvement in linearity, relative to the rawspecies abundances. However, the deviation from linearity atthe extremes of the distributions in the plot indicate thatthe data were not multivariate normal. The failure of thedata to meet the assumption of multinormality has removed theopportunity for formal hypothesis tests and attachment ofprobabilities to the results of the analyses. The percent ofvariance accounted for by the various ordinations indicatesthe strength of the conclusions arrived at via the analyses.It follows that the results should be interpreted with somereservation.19Figure 1. Number and proportion of sea urchin test diameters(cm) collected from Bath Island in 1992.202015 1510oa 50a) b)^ • •505^10^15^20^25^30^5^10^15^20^25^30EXPECTED VALUE EXPECTED VALUEFigure 2. Q-Q probability plots of a) raw species abundancesand b) log-transformed species abundances.Distance measure isMahalanobis distance from group centroid, expected valuebased on Chi-squared distribution (32 degrees of freedom).21Principal Components AnalysisThe PCA of the first log-transformed data subset (allquadrats, species occurring more than 15 times) yielded 66component axes, the most important of which accounted foronly 7.6% of the variation in the data set (Table 2). Thecomponent loadings from the first three axes resulting fromthis analysis are included in Table 2, but will not beinterpreted further as the following analysis summarizes themain points in a more concise manner.The PCA of the second log-transformed data subset(quadrats between 3 m above and 6 m below zero tide, specieswith high component loadings on one of the first threecomponent axes from the first PCA) yielded 27 component axes,the first three accounting for 23.2%, 9.7%, and 8.3% of thevariation in the data subset (Table 3). The componentloadings on the first axis indicate a number of species withstrong, positive correlations, and others with strongnegative correlations. Those species with strong positivecorrelations include Laurencia spectablis (0.732), Plocamiumcartilagineum (0.727), Prionitis spp. (0.703), Iridaeasplendens (0.781), and Constantinea subulifera (0.779).Strong negative correlations occur with Fucus spp. (-0.405)and Mastocarpus papillatus (-0.419) (Table 3).The second axis also indicates groups of positively andnegatively correlating species. Strong positive correlationsare found between the second axis and Gelidium spp. (0.443),22Table 2.^Eigenvalues, percent variance accounted for,component loadings for the first three axes from theprincipal components analysis of raw species abundancesall quadrats.andinAxis 1 Axis 2 Axis 3Eigenvalue 5.044 2.702 2.314% Variance 7.643 4.094 3.506Species Component LoadingsLaurencia spectablis 0.367 -0.312 -0.116Herposiphonia plumula 0.750 0.172 0.055Pterosiphonia dendroidea 0.444 0.188 -0.151Amplisiphonia Pacifica 0.460 0.096 0.164Ulvaria obscura 0.108 -0.184 0.067Plocamium cartilagineum 0.694 -0.042 0.288Corallina officinalis 0.175 -0.490 0.007Odonthalia floccosa 0.100 0.019 -0.150Ceramium pacificum 0.371 -0.401 0.081Constantinea subulifera 0.458 0.045 -0.509Prionitis spp. 0.309 -0.427 -0.112Iridaea splendens 0.389 -0.071 -0.369Antithamnion defectum 0.478 0.233 0.005Gelidium spp. 0.024 -0.547 0.006Pterosiphonia gracilis 0.045 0.172 0.042Ulva spp. -0.040 -0.138 0.014Microcladia borealis -0.006 -0.114 -0.128Fucus spp. -0.107 0.096 0.020Polysiphonia spp. 0.016 0.122 -0.025Bossiella chiloensis -0.022 -0.088 0.044Calliarthron tuberculosum 0.011 0.008 -0.002Callophyllis flabellulata 0.691 0.091 -0.229Gymnogrongus leptophyllus 0.016 -0.332 -0.087Cryptopleura ruprechtiana 0.592 -0.036 -0.005Polyneura latissima 0.605 0.061 0.517Platythamnion pectinatum 0.014 0.084 0.061Sargassum muticum 0.003 -0.586 0.016Bossiella spp. 0.003 -0.108 -0.003Mastocarpus papillatus -0.104 0.084 0.025Agarum fimbriatum -0.061 0.070 0.045Laminaria saccharina 0.113 -0.009 -0.134Desmarestia viridis 0.205 0.032 -0.216Bonnemaisonia nootkana 0.004 0.141 0.09023Colonial Diatoms 0.003 -0.045 -0.042Callophyllis heanophylla 0.056 0.006 -0.090Nereocystis luetkeana -0.005 0.034 -0.052Neorhodomela larix -0.042 -0.189 -0.073Branchioglossum bipinnatifidum 0.009 -0.384 0.039Colpomenia peregrina -0.025 -0.166 -0.004Costaria costata -0.009 0.009 0.030Rhodoptilum plumosum -0.038 0.059 0.028Gigartina exasperata 0.101 -0.482 0.111Enteromorpha spp. -0.044 0.033 0.016Leathesia difformis -0.026 -0.052 0.011Rhodymenia pertusa 0.312 0.161 -0.503Platythamnion villosum 0.143 0.190 -0.041Rhodoglossum roseum 0.217 -0.015 -0.226Pugetia fragilissima -0.057 -0.011 0.031Lomentaria hakodatensis -0.027 -0.417 0.041Ceramium californicum 0.009 0.104 0.094Laminaria bongardiana 0.249 0.130 -0.395Hollenbergia subulata 0.027 0.173 0.190Porphyra spp. -0.060 0.077 0.024Sphacelaria spp. 0.014 0.007 0.014Heterosiphonia densuiscula -0.025 0.202 0.177Callophyllis violacea 0.594 -0.050 -0.227Schizymenia pacifica 0.090 -0.012 -0.073Nienburgia andersoniana 0.315 0.136 -0.166P1eonosporium vancouveranium 0.055 -0.025 0.019Laminaria farlowii -0.020 -0.341 -0.036Microcladia coulteri 0.534 0.023 0.559Fauchea spp. -0.039 -0.046 0.115Cryptosiphonia woodii -0.014 0.016 0.002Hara1diophyl1um mirable 0.540 0.037 0.554Weeksia coccinea -0.039 0.057 0.039Cladophora spp. -0.023 -0.032 0.02824Table 3. PCA of log-transformed species data. Onlyeigenvalues >2 shown.^Axis 1^Axis 2^Axis 3Eigenvalue^ 6.257^2.622^2.230% Variance 23.174 9.710 8.261Species^ Component LoadingsLaurencia spectablisPlocamium cartilagineumOdonthalia floccosaCorallina officinalisCeramium pacificumPrionitis spp.Iridaea splendensConstantinea subuliferaUlvaria obscuraLaminaria spp.Ulva spp.Fucus spp.Gelidium spp.Callophydlis spp.Calliarthron tuberculosumCryptopleura ruprechtianaSargassum muticumMastocarpus papillatusDesmarestia viridisNeorhodomela larixPolyneura latissimaGigartina exasperataLomentaria hakodatensisPorphyra spp.Agarum fimbriatumCostaria costataNereocystis luetkeana^0.732^0.013^-0.0310.727 0.020 0.1210.587^-0.042^0.2220.612 0.339^-0.1360.587^0.252 0.3220.703 0.040^0.3270.781^-0.039 0.2610.779^-0.057^0.1740.294^-0.112^-0.0900.442^-0.039^-0.5040.063 0.472 0.275-0.405^0.208^0.5530.434^0.443^-0.4180.674^-0.450^0.1970.118^-0.256^-0.1790.619^-0.387 0.3120.388 0.726^-0.050-0.419^0.156^0.4580.363^-0.427 0.018-0.041 0.287^0.2550.352^-0.218 0.1260.396^0.462^-0.2800.239 0.557^-0.110-0.176^0.048 0.2690.009^0.000^-0.2410.221^-0.070^-0.4820.228^-0.382^-0.32525Ulva spp.(0.472), Sargassum muticum (0.726), Gigartinaexasperata (0.462), and Lomentaria hakodatensis (0.557),while strong negative correlations are found with Nereocystisluetkeana (-0.382) and Cryptopleura ruprechtiana (-0.387).Axis 3 shows a positive correlation with Ceramiumpacificum (0.322), Prionitis spp. (0.327), Fucus spp.(0.553), Cryptopleura ruprechtiana (0.312), and Mastocarpuspapillatus (0.458). A negative correlation exists withLaminaria spp. (-0.504), Gelidium spp. (-0.418), Costariacostata (-0.482), and Nereocystis luetkeana (-0.325).Figures 3 and 4 show the relationship between quadratdepth and quadrat scores on the first and second componentaxes. Low scoring quadrats on the first axis apparentlyoccur above 0 tide level to 3 m, intertidally (Fig. 3),corresponding to the Fucus spp. and M. papillatusassociation. The first axis therefore seems to separateintertidal quadrats from subtidal ones. On the second axis(Fig. 4), both high scoring and low scoring quadrats arepresent deeper than 0 tide level. However, virtually all ofthe high scoring sites were sampled in 1992, while the lowscoring sites appear to have been sampled exclusively duringthe 1970's. Above 0 tide level to 3 m, high scores arepresent for quadrats sampled from both time periods,indicating that a second intertidal macrophyte association ispresent. The second axis seems to describe both an Ulva spp.dominated intertidal association and a 1992 subtidal26Figure 3. Quadrat scores on the first principal axis fromthe PCA of log-transformed species abundances vs. quadratdepth (m). Negative values for depth indicate depth belowzero tide level.27Figure 4. Quadrat scores on the second principal axis fromthe PCA of log-transformed species abundances vs. quadratdepth (m); quadrats sampled in 1992 (0) and during the 1970's(•). Negative values for depth indicate depth below zerotide level.2 8association at the positive end, and a 1970's subtidalassociation at the negative end.Figures 5a through 5i show the quadrat scores for eachsite/year on the first two component axes, with 95%confidence intervals indicated by the ellipses. Table 4summarizes the changes in area (A), angle of orientation (0),and eccentricity (ecc) of the ellipses that are apparent byvisual inspection of Figures 5a through 5i. At Bath Island,a decrease in area and an increase in eccentricity from 1972to 1973 corresponds to the year the urchins arrived at thesite (Figs. 5a, 5b; Table 4). The decrease in area seems toresult from a decrease in quadrat scores at the positive endof axis 1, and from a decrease at the negative end of axis 2.From 1973 through 1975, the area is increasing, apparentlydue to the decreasing eccentricity values which indicatelower correlations among species. In 1992, Bath Island hassimilar area and eccentricity to 1974 and 1975, but differsgreatly in angle of orientation (Figs. 5b through 5e; Table4). The Sear Island site/years tend to have similar anglesof orientation to the corresponding Bath Island site/years,but have much greater values for their area and eccentricity.The low angles of orientation of the two 1992 site/years aredramatically different from those of the 1970's, suggestingthat the 1992 site/years are similar to each other in termsof their species compositions. However, in terms of area(within site/year variation) and eccentricity (the extent towhich species associations are correlated), the 1992295 53 31 1Noa_^13 3-5-5^3^-1^1PC(1)3-53^5^-5^3^-1^1PC(1)553 31 13 33 5 3 55-5-5^3^-1^1PC(1)-5-5^3^-1^1PC(1)31a(.30_^13-5-5^3^-1^1PC(1)3 5305 53 31 13 3-5-5^3^-1^1PC(1)-53 5 -5 -1^1PC(1 )53313-53 55^3^-1^1PC(1)3 55313-5-5^3^-1^1PC(1)Figure 5. Quadrat scores on second principal axis vs. scoreson first principal axis from the PCA of log-transformedspecies abundances (all quadrats between 3 m above and 6 mbelow zero tide). Ellipses represent 95% confidenceintervals; a) B72, b) B73, c) B74, d) B75, e) B92, f) S72, g)S74, h) S75, i) S92.31TABLE 4. Statistics describing the 95% confidence ellipsesfrom PCA of log-transformed data. 0, angle of major axis withx axis; ecc, eccentricity. See text for explanation ofdescriptive statistics.Site/Year^ecc^0^AreaB72^.893^145^565.5B73 .951 148 205.0B74 .706 108 320.4B75^.650^164^373.1B92 .768 38 314.2S72 .866 145 1061.9S74^.932^162^552.9S75 .958 164 395.8S92 .973 53 490.1 32site/years are more similar to the site/years from theirrespective islands, particularly in 1974 and 1975, than toeach other.Canonical Variates AnalysisFormal multivariate tests of the equality of site/yearcommunity centroids and the dimensionality of canonical spacecould not be applied based on the failure of the data set tomeet the assumption of multinormality. The first threecanonical variates were thus chosen for interpretation basedonly on the strength of the canonical correlations (0.709,0.680, 0.560; Table 5).The correlation coefficients between the canonicalvariates, uk and xrk , and the variables in the dummy andspecies variable sets are shown in Table 5. Thesecorrelation coefficients will be used for interpretation ofthe canonical variates.Species that show a relatively strong positivecorrelation with vi include Ulvaria obscura (0.357),Callophyllis spp.(0.563), Cryptopleura ruprechtiana(0.538), Desmarestia viridis (0.311), Polyneura latissima(0.330), and Nereocystis luetkeana (0.414). Those speciesshowing relatively strong negative correlations with vi areSargassum muticum (-0.419), Corallina officinalis (-0.284),Gelidium spp. (-0.329), Ulva spp. (-0.432), and Gigartinaexasperata (-0.276). This variate extracts 7.56% of the33TABLE 5. Canonical correlation coefficients, redundancies, %variance extracted, and canonical loadings from the CVA oflog-transformed species abundances.Correlation Coefficient^0.709^0.680^0.560Canonical Variate^ vl^v2^v3Species^ Canonical Loadings-0.009 0.289 0.0380.108 0.427 0.0120.053 0.146 -0.225-0.284 0.510 0.134-0.002 0.171 0.1220.096 0.309 0.1270.224 0.363 0.1240.184 0.476 0.0590.357 -0.258 0.3480.070 0.481 -0.032-0.432 0.068 0.1940.104 -0.096 0.021-0.329 0.577 0.1700.563 0.338 0.065-0.090 -0.275 -0.2000.538 0.254 0.119-0.419 0.319 0.1300.183 0.080 0.0740.311 0.230 -0.472-0.177 -0.011 -0.1040.330 0.161 0.095-0.276 0.396 0.453-0.272 0.326 0.2810.153 -0.126 0.106-0.079 0.208 0.0380.046 0.267 -0.2540.414 0.217 -0.2687.560 9.470 3.9403.800 4.380 1.240Laurencia spectablisPlocamium cartilagineumOdonthalia floccosaCorallina officinalisCeramium pacificumPrionitis spp.Iridaea splendensConstantinea subuliferaUlvaria obscuraLaminaria spp.Ulva spp.Fucus spp.Gelidium spp.Callophyllis spp.Calliarthron tuberculosumCryptopleura ruprechtianaSargassum muticumMastocarpus papillatusDesmarestia viridisNeorhodomela larixPolyneura latissimaGigartina exasperataLomentaria hakodatensisPorphyra spp.Agarum fimbriatumCostaria costataNereocystis luetkeana% VarianceRedundancy34Canonical Variate ul u2 u3Canonical LoadingsB73 -0.073 0.278 -0.215B74 -0.101 0.571 0.261B75 -0.066 0.137 -0.082B92 0.454 0.379 -0.160S72 -0.477 -0.436 0.182S74 -0.337 0.012 0.615S75 -0.128 -0.029 0.060S92 0.697 -0.562 0.265% Variance 13.400 13.400 7.900Redundancy 6.720 6.200 2.48035variance in the species data set and 3.8% of the variation inthe site/year data (Table 5). The site/year communities thatcorrelate with u l are B92 (0.454) and S92 (0.697) in apositive manner, and S72 (-0.477) and S74 (-0.337) in astrong negative manner. The u 1 variate extracts 13.4% of thevariation in the site/year data, and accounts for 6.72% ofthe variation in the species data (Table 5). Therelationship between vl and u l is shown graphically in Figure6a. From this it can be seen that the high scores of S92 andB92 on u l correspond to their low scores on v1 . Thesite/years S92 and B92 differ from the rest of the site/yearcommunities due to a high proportion of S. muticum, C.officinalis, Gelidium spp., Ulva spp., and G. exasperata.The v2 variate correlates positively with species suchas Constantinea subulifera (0.476), Laminaria spp. (0.481),Gelidium spp. (0.577), Plocamium cartilagineum (0.427), andCorallina officinalis (0.510). Negatively correlated speciesinclude Ulvaria obscura (-0.258) and Calliarthrontuberculosum (-0.275) (Table 5). The second variate extracts9.47% of the variance in the species data. The u 2 variatecorrelates positively with B73, B74, B75 and B92 andnegatively with S72, S75, and S92. Figure 6b shows that thesites correlating positively with u 2 have low scores on v2 ,corresponding to higher amounts of U. obscura and C.tuberculosum, while the other site/years have greater amountsof C. subulifera, Laminaria spp., Gelidium spp., P.cartilagineum, and C. officinalis.36a)2 tere► tr 31611A41111P80,111141111111o d A tal&III■ MA AI AY h.0 0 • ^0 00Cam 115•011 •••••13 •• ■ CO^CID DD b MIN NNND tn. NW&-3^-2^-1^0^1^2^3Vi373 81m 0—1—2 oa5o^o (lumpIZPP^DOCIPtit► t, MK>^MPIIMINIIIMMIANAltnilIMINCr re cla *0111■0 CO 0c"^16l90 0 0 oammocco▪ ow m▪^mum •1^1^1^i^1^1 —4 —3 —2 —1 0^1^2^3^4V3Figure 6. Quadrat scores on vi and ui axes from the CVA oflog-transformed species abundances; a) u 1 vs. v1 , b) u2 vs.v2 , c) u 3 vs. v3 ( • B72, 0 B73, - B74, B75, I B92, ° S72,0 S74,^S75,^S92).39Species correlating with the third variate, v3 , areUlvaria obscura (0.348) and Gigartina exasperata (0.453)positively, and Odonthalia floccosa (-0.225), Calliarthrontuberculosum (-0.200), Desmarestia viridis (-0.472), Costariacostata (-0.254), and Nereocystis luetkeana (-0.268)negatively. The v 3 variate extracts 3.94% of the variance inthe species data (Table 5). Figure 6c indicates that the lowscoring sites on u 3 have lower scores on v3, corresponding togreater amounts of 0. floccosa, C. tuberculosum, D. viridis,C. costata, and N. luetkeana.Figure 7a represents a three dimensional scatter plot of thequadrats on the ui axes. Since the ui scores of the quadrats froma particular site/year are identical, the quadrats from eachsite/year appear as a single point on the graph. Further, the u iscores of the quadrats from a particular site/year are correlatedto their scores on the corresponding vi axis, so site/years withsimilar species compositions should have similar locations on theui axes, and thus cluster together in Figure 7a. Figures 7b and7c are two dimensional representations of Figure 7a, to aid invisualization of the three dimensions. The site/years B73 and B75appear close on all three axes, with B92 different from them onlyon ul Site/year B74 is somewhat similar but differs on u 2  Thesite/years S74 and S75 appear to be similar to each other, withmost of their separation occurring on u 3 . The other site/yearsseem to be unique, sharing similar scores with other site/years ononly one axis. For example, B92 and S92 share a highUI-4^0-2-121b)B92^ S92B74B73. 9 vS75 oS74^ B72p S72c)^a B74B73 ooi^ S74 ^S75B72t. S72S9221ry^0-1-22^1^0^-1^-2^-2^-1^0^1^2U2 U3Figure 7. Scatter plots of quadrat scores on u l , u2 and u 3 fromthe CVA of log-transformed species data; a) three dimensionalplot, b) u l vs u2 , c) u 2 vs u3 . b) and c) represent twodimensional views of a) along the u 3 axis and ul axis respectively(• B72, ° B73, - B74, - B75, ° B92,^S72, ^ S74, ° S75,^S92).separation from the other sites along the first axis, but arevery different on u 2 and u3 .Canonical Correlations AnalysisThis analysis correlates the same data sets as the CVA,but includes depth as an additional variable to the dummyvariable set. The first three canonical correlations areshown in Table 6. The canonical loadings on the vi variatefrom the strongest correlation (0.784), indicate that Fucusspp. (0.737), Mastocarpus papillatus (0.607), Porphyraspp.(0.315), and Neorhodomela larix (0.232) are strongpositive contributors to that correlation, and Laminaria spp.(-0.525), Gelidium spp.(-0.476), Agarum fimbriatum (-0.461),and Corallina officinalis (-0.399), among others, are strongnegative contributors. Figure 8 shows the strong effect thatdepth has on the quadrat scores, with shallow quadratsscoring much higher on vl (corresponding to the intertidalspecies Fucus spp., M. papillatus, Porphyra spp., and N.larix) than the deep quadrats. Figure 9a shows that most ofthe separation between quadrats on u l is due to depth, whilevery little separation between site/years is present.The canonical loadings (Table 6) on v2 show thatCallophyllis spp.(0.649), Cryptopleura ruprechtiana(0.593), Desmarestia viridis (0.389), Polyneura latissima(0.373), and Nereocystis luetkeana (0.508) are42TABLE 6. Canonical correlation coefficients, redundancies, %variance extracted, and canonical loadings from the CCO oflog-transformed species abundances.Correlation Coefficient^.784^.704^.655Canonical Variate^ vl^v2^v3Species^ Canonical Loadings-0.324 0.133-0.201 0.257-0.011 0.093-0.399 -0.0680.010 0.0390.014 0.168-0.078 0.330-0.190 0.3420.107 0.263-0.525 0.3070.141 -0.4460.737 -0.082-0.476 -0.078-0.040 0.649-0.139 -0.1300.024 0.593-0.183 -0.2950.607 0.067-0.093 0.3890.232 -0.230-0.033 0.373-0.354 -0.097-0.171 -0.1520.315 0.050-0.461 0.077-0.333 0.187-0.208 0.5089.370 8.4705.760 4.200Laurencia spectablisPlocamium cartilagineumOdonthalia floccosaCorallina officinalisCeramium pacificumPrionitis spp.Iridaea splendensConstantinea subuliferaUlvaria obscuraLaminaria spp.Ulva spp.Fucus spp.Gelidium spp.Callophyllis spp.Calliarthron tuberculosumCryptopleura ruprechtianaSargassum muticumMastocarpus papillatusDesmarestia viridisNeorhodomela larixPolyneura latissimaGigartina exasperataLomentaria hakodatensisPorphyra spp.Agarum fimbriatumCostaria costataNereocystis luetkeana% VarianceRedundancy0.1360.3100.1210.4470.1940.3210.2610.339-0.3450.1710.3590.2970.4850.134-0.3350.0720.4070.3840.0540.2030.0310.3400.373-0.004-0.0270.066-0.0447.4403.19043Canonical Variate ul u2 u3Canonical LoadingsB73 0.161 0.032 0.242B74 0.227 0.095 0.518B75 -0.036 -0.044 0.220B92 0.109 0.576 0.133S72 0.114 -0.549 -0.398S74 0.115 -0.305 0.072S75 -0.050 -0.151 0.076S92 -0.451 0.449 -0.619Depth -0.866 -0.209 0.356% Variance 11.920 11.180 11.910Redundancy 7.320 5.540 5.11044••▪ ••I•• .e• •.:• 1.06! I : . 11 •• •• a l • .• I• a • •••• • •^•r• I •^ir• ;: I r^.■•1111:1:11• 1."• •^• •..•1. •^• •• • • • • 1.••••6^3^0^-3^-6^-9Depth (m)Figure 8. Quadrat scores on vi from the canonicalcorrelations analysis of log-transformed species abundances,site/year dummy variables and depth vs. quadrat depth (m).Negative values for depth indicate depths below zero tidelevel.- 1- 2- 3-445positively correlated, whereas Sargassum muticum (-0.249),Calliarthron tuberculosum (-0.130), Ulva spp. (-0.446), andNeorhodomela larix (-0.230) are negatively correlated.Figure 9b shows that B92 and S92 score higher than the othersite/years on u2 . The relationship between v2 and u2 (Figs.10a, 10b) indicates that low scores on v2 account for thehigh scores of B92 and S92 on u 2 . The separation of B92 andS92 from the other sites is apparently due to higherabundances of S. muticum, C. tuberculosum, Ulva spp., andN. larix.The relationships between v3 and Constantinea subulifera(0.339), Gelidium spp. (0.485), Plocamium cartilagineum(0.310), and Corallina officinalis (0.447) are among the manythat are strong and positive, whereas negatively correlatedspecies include Ulvaria obscura (-0.345), and Calliarthrontuberculosum (-0.335) (Table 6). The scores on u 3 indicate aseparation between low scoring site/years such as S92, S72and B72 (none of which were grazed by sea urchins), andhigher scoring site/years such as B92, B73 and B74 (all ofwhich were grazed by sea urchins)(Fig. 9c). Figure lla andllb indicate that the low scores of S92, S72, and B72 on u 3are mainly accounted for by high scores on v3 , or highabundances of such species as C. subulifera, Gelidium spp.,P. cartilagineum, and C. officinalis. The high scoringsite/years on u 3 likewise appear to have more U. obscura andC. tuberculosum. These site/year groupings are similar tothose that were indicated by the second canonical correlation46of the CVA analysis.470a)FEB•r4 •** •10-1- 26^3^0^-3Depth (m)48Depth (m)4 9Figure 9. Quadrat scores on u i from the canonicalcorrelations analysis of log-transformed species abundances,site/year dummy variables and depth vs. quadrat depth (m); a)ul from first correlation, b) u 2 from second correlation, c)u3 from third correlation (• B72, o B73, ^ B74, ■ B75,^B92,S72,^S74, • S75, lb S92). Negative values for depthindicate depth below zero tide level.5 05 1C.410-1-2-3 b)4noikurtr * **^41,4lw -Lb^ ^,oc,0^c'P 06Ep 2^0,_13El OD 0M00 1:1to^tbp^P P14eP^-3^-2^0^1^2^3^4V2Figure 10. Quadrat scores on u2 and v2 from the canonicalcorrelations analysis of the log-transformed speciesabundances, site/year dummy variables and depth. Bath Islandquadrats (a) and Sear Island quadrats (b) are shown onseparate graphs to minimize distortion caused by overlappingpoints (• B72,^° B73, - B74, -B75,^B92, b S72, 0 S74,0 S75,^S92).5253Aaa-2 -1 0V31 2 332m 10a)Ao v voo f; oci00■• ••I • I■0Figure 11. Quadrat scores on u 3 and v3 from the canonicalcorrelations analysis of the log-transformed speciesabundances, site/year dummy variables and depth. Bath Islandquadrats (a) and Sear Island quadrats (b) are shown onseparate graphs to minimize distortion caused by overlappingpoints ( • B72, ° B73, - B74, - B75, 4 B92,^S72, 0 S74,° S75, 4, S92).5 4DISCUSSIONThe objectives of this study have been to show changesin macrophyte community structure 1) along the depthgradient, 2) in response to grazing, 3) after removal ofgrazing pressure, and 4) over the long term from the 1970'sto 1992. Although formal tests of hypotheses were notpossible based on the failure of the data to meet theassumption of multinormality, much information regardingcommunity change has been learned from'the analyses. Foreach objective, various aspects of the analyses combine toidentify macrophyte associations which are indicative ofcommunity change.Community Change Along the Depth GradientThe PCA and CCO included the effects of depth on thespecies associations, and in both analyses, the first axesrevealed a separation between intertidal and subtidalassociations. In fact, the component loadings on the firstprincipal component axis and the vi axis from the CCO arehighly similar, (the sign of the correlation is arbitrary)showing the same trends for virtually all of the species.The species identified with the intertidal association areFucus spp., Mastocarpus papillatus, Neorhodomela larix, andPorphyra spp. Ulva spp. may also be associated with thesespecies, as it has a somewhat similar loading on the vi axis.55from the CCO. Further, Ulva spp. appears to correlate withthe second principal component analysis in a similar manneras N. larix, and based on the relationship between depth andthe second principal component axis, Ulva spp. is probablycontributing to the high scores of the intertidal quadrats onthis axis. Fucus spp., M. papillatus, and N. larix werepreviously identified as characteristic species of anintertidal algal community at the same sites (Lindstrom andForeman 1978), using a species-quadrat coincidence tablebased on the 1972 data only.The other species on the first axes of the PCA and CCOseem to comprise a widely distributed subtidal association offoliose red algae and kelp species. Most of these specieswere included as characteristic species of either the foliosered algal community (i.e. Plocamium cartilagineum,Constantinea subulifera, Prionitis spp.) or the kelpcommunity (i.e. Laurencia spectablis, Corallina officinalis,Laminaria spp., Costaria costata) described by Lindstrom andForeman (1978). Although there was no separation along thedepth gradient evident between these communities on the firstaxes, the third axis of the PCA may indicate a separationbetween quadrats in the foliose red algal community and thekelp community, which would be more consistent with thefindings of Lindstrom and Foreman (1978).56Community Change and Sea Urchin Grazing ActivityThe subtidal association evident from the first axes ofthe PCA and CCO analyses also yielded information on theeffects of grazing on community structure. The kelp/foliosered association was apparently that which was most stronglyimpacted by the grazing activity of the urchins. The lowscores of the Bath Island site (Figs. 5a through 5e) at thepositive end of the first principal component axis in.1973and 1992, years in which urchins were present at the site,and 1974 and 1975, the two years directly following theurchin removal, indicate a low abundance of speciescomprising this association. The high scores of the SearIsland quadrats, free of urchins during all years ofsampling, at the positive end of the axis (Fig. 5f through5i) contrasts with these Bath Island site/years. Further,the decrease in area of the ellipses at Bath Island from 1972to 1973 (Fig. 5a, 5b; Table 4) indicates an overall decreasein the variability of species assemblages in 1973.This decrease in species variability is consistent withother observations of decreased macroalgal abundance inresponse to heavy urchin grazing (Foreman 1977, Duggins 1980,Dean et al. 1984, Ebeling et al. 1985) and with traditionalviews of intensive herbivory in terrestrial plant communities(Crawley 1983).Elements of the kelp/foliose red subtidal associationwere also apparent in the canonical loadings of v2 from the57CVA, and v3 from the CCO. These axes were highly similar,and seemed to separate quadrats of the kelp/foliose redassociation from those which contained a large amount ofCalliarthron tuberculosum and Ulvaria obscura. Quadrats thatwere sampled at Bath Island following the arrival of theurchins scored low on these axes relative to those that werenot (Fig. 9b). This suggests that the C. tuberculosumassociation exists as a grazed form of the subtidalkelp/foliose red association. Coralline algae such as C.tuberculosum have consistently been found to persist inheavily grazed "barrens" areas described in other studies ofsea urchin grazing (Pearse and Hines 1979, Harrold and Reed1985). Further, Paine and Vadas (1969) found Ulva-Gayraliaspecies were the only foliose algae to consistently tolerategrazing in enclosed cages with sea urchins.This evidence suggests that the grazing activity by seaurchins effectively reduced the overall variability of themacroalgal assemblages by removing various foliose red andkelp species. Grazing resistant species such as Calliarthrontuberculosum and Ulvaria obscura were not affected by the seaurchins and so remained in disproportionate abundancefollowing grazing. These findings are in accordance withthose described by Foreman (1977), who found, based on aspecies importance measure (frequency x mean biomass), thatthe most important species during 1973 were members of theChlorophyceae and encrusting or articulated corallinespecies. Foreman (1977) postulated that the persistence of58the Chlorophyceae may result from their weedy nature. Theremoval of individuals by grazing would have little influenceon the population which would quickly re-establish, andperhaps benefit by the reduced abundance of potentialcompetitors.Community Change After Sea Urchin RemovalThe extent of recovery of the macrophyte communities atBath Island after removal of the sea urchins in 1973 wassummarized by the PCA, evident in Figures 5b through 5d.There is an increasing trend in the ellipse areas from 1973through 1975, corresponding to increased variability amongquadrats of each site/year. The trend contrasts that whichis evident at Sear Island over the same time period (Fig. 5g,5h), which lends to the supposition that community enrichmentis occurring. The change in area occurs with a decrease inthe value for 0 in 1974 (Fig. 5c), reflecting the higherquadrat scores at the positive end of PC2. This correspondsto small increases in Ulva spp., Sargassum muticum,Lomentaria hakodatensis, Gigartina exasperata, and Gelidiumspp. S. muticum has been shown to colonize experimentallydenuded surfaces at Bath Island, particularly in summermonths (DeWreede 1983). As S. muticum is the strongestcontributor to PC2, the higher scores of B74 on this axisprobably reflect the recruitment of S. muticum in the freespace created by the sea urchins in 1973. A further59contributor to the higher scores of B74 on PC2 is Ulva spp.,which was also observed to colonize experimentally createdfree space by DeWreede (1983).The increase in abundance of these species did notcontinue in 1975 where 0 has increased (Table 4),corresponding to decreased scores at the positive end of PC2(Fig. 5d). This is consistent with the findings of DeWreede(1983), who found poor over-winter survival of Sargassummuticum that had colonized free space in the summer of 1976and 1978. Concurrent with the decrease in the speciesindicated by PC2, the slight increase in the scores of B75 atthe positive end of PC1 indicates increased abundances offoliose red and kelp species in the second year afterdisturbance.At Bath Island from 1973 through 1975, the trends incommunity structure indicated by the positive end of PC1 andPC2 would suggest a change from the low variability and heavyimpact described for 1973, towards an increased abundance ofspace colonizing species such as Sargassum muticum and Ulvaspp. in 1974. By 1975, increasing trends in ellipse area, aswell as decreased scores at the positive end of PC2 andincreased scores at the positive end of PC1 suggest a trendin overall macrophyte composition of B75 towards that of B72.However, the sample size of B75 is small and conclusionsshould be viewed with caution.The descriptive methods of interpretation employed byForeman (1977), based on measures of species diversity,60abundance of 10 arbitrary growth form/taxonomic macrophytegroups, and a species importance measure (frequency x meanbiomass), indicated similar trends in communityregeneration. In 1974, foliose reds and annual kelps (i.e.Nereocystis luetkeana) increased in importance and relativeabundance. Also noted was a maximum importance value forSargassum muticum, which was postulated to represent theearly foliose red community. The return of pre-disturbanceimportance levels for Constantinea subulifera, Plocamiumcartilagineum and Iridaea splendens was apparent in 1975.Although an increase in 1975 for these species was suggestedby the multivariate analyses, there was no indication thatthey had reached their 1972 pre-grazing abundance. Thissuggests that if detailed information on the response ofparticular species was required in a study, univariateanalyses by individual species would be more appropriate.The same conclusion has been reached by other researchersusing a combination of multivariate and univariate approachesto successional change in plant communities (Gibson and Brown1992).Community Change from the 1970's to 1992Another association that appeared consistently in allthree of the analyses was a shallow subtidal association ofSargassum muticum, Gigartina exasperata, Lomentariahakodatensis, Corallina officinalis, Gelidium spp., and Viva61spp. High scores on axis 2 from the second PCA, low scoreson vi from the CVA, and low scores on v2 from the CCO allindicated this association. The relationship between thisassociation and the 1992 quadrats from both Bath and SearIsland (Figs. 4, 5i, 7b, 9b, 10a, and 10b) suggests thatthese species have increased in abundance at the two sitesfrom the 1970's through 1992. Of these species, onlyGelidium spp. was identified as characteristic of one of thecommunities discussed by Lindstrom and Foreman (1978),further indicating that a redistribution of species hasoccurred at the sites since the 1970's.The 1992 shallow subtidal association contrasts withanother association of species including Callophyllis spp.,Cryptopleura ruprechtiana, Nereocystis luetkeana, Polyneuralatissima, and Desmarestia viridis. These species werecorrelated positively with v l from the CVA and v2 from theCCO, and negatively with the second principal component axis.The relationship between the quadrats sampled during the1970's and these species is apparent from the same evidenceas that which identified the 1992 shallow subtidalassociation (Figs. 4, 5i, 7b, 9b, 10a, and 10b). It appearsthat sites sampled during the 1970's accounted for virtuallyall of the negatively scoring quadrats on the secondprincipal component axis (Fig. 4), particularly during 1972at Bath and Sear Island (Figs. 5a, 5f). Because of theoverlapping depth distribution of positively and negativelyscoring quadrats on this axis (Fig. 4), Callophyllis spp.,62C. ruprechtiana, P. latissima, and D. viridis represent anassociation that has been replaced by the 1992 shallowsubtidal association. The differences in upper subtidalmacrophyte community structure between the early 1970's and1992 seems to be summarized by increases in abundance ofSargassum muticum, Gigartina exasperata, Lomentariahakodatensis, Corallina officinalis, Gelidium spp., and Ulvaspp., and decreases in Callophyllis spp., C. ruprechtiana, N.luetkeana, P. latissima, and D. viridis.Sargassum muticum was introduced to the northeastPacific from Japanese oyster spat in the 1930's or 1940's(Scagel 1956), and has since spread as far south as BajaCalifornia and as far north as Southeast Alaska (Lindstrom1977). A similar origin in the northeast Pacific ispostulated for Lomentaria hakodatensis, which has also sincespread south in its distribution (Hawkes and Scagel 1986)The increases of these two species apparent in this study isindicative of further success of these introduced species inthe northeast Pacific.Studies of community change or constancy in benthicmarine communities have primarily dealt with changes in smallpatches, involving recolonization after a disturbance event,or the question of succession (Connell 1987). This focus,emphasized during the investigation of the communityresponses to grazing in the first part of this study,contrasts with the more long-term changes that are suggestedhere. The strongest evidence that these changes represent63long-term changes in community structure as opposed to yearlyrandom variation in species abundances is as follows. Werethe differences in community structure between the 1992site/years and the 1970's site/years a result of annualvariation in recruitment, it is likely that other site/yearswould have shown similar community structures, high inabundance of Sargassum muticum, Gigartina exasperata, andUlva spp. and low in abundance of Callophyllis spp.,Nereocystis luetkeana, Polyneura latissima, and Desmarestiaviridis. As it was, only the 1992 site/years showed thistrend. However, continuous sampling efforts would bemandatory to distinguish between annual variation incommunity structure and long-term changes. Further, todetermine if the changes were widespread throughout theStrait of Georgia, more sites at random locations throughoutthe strait would need to be sampled. A continuous data setfor these sites could have provided more information on thecauses of community change, and suggested more specific areasin which to concentrate further experimental studies.Community changes that have occurred over long periodsof time may result from gradual shifts in local physicalconditions (Connell 1987). One possible explanation for theobserved shift in community composition from the 1970's to1992 may be related to the temperature tolerance of thevarious species involved in the changes. In a study oftemperature tolerance among northeast Pacific macroalgae,Sargassum muticum, Gigartina exasperata, and Ulva fenestrata64were among the most tolerant species tested for survival athigh temperatures whereas Callophyllis spp., Nereocystisluetkeana, Polyneura latissima, and Desmarestia viridis allappeared in lower temperature tolerant groups (LUning andFreshwater 1988). By inspection of monthly mean surfacetemperatures at Active Pass, near the study sites (Fig. 12),it appears that a gradual increase of perhaps 2 °C in averagetemperature at the study sites has occured. There is thusthe possibility that temperature changes in the strait havecreated an environment in 1992 that is favourable to adifferent pool of species (S. muticum, G. exasperata, andUlva spp.) from that in the 1970's (favourable toCallophyllis spp., N. luetkeana, P. latissima, and D.viridis)The above postulation is suggested with much reserve,as other explanations may well describe the differences inabundance of the above species. For example, the lowabundance of Nereocystis luetkeana in the 1992 quadrats mayreflect the state of recovery of the sites from grazing atthe time of sampling. N. luetkeana has been shown to behaveas a "fugitive species", establishing itself at grazed sitessoon after removal of the urchins, and gradually beingreplaced by perennial kelp species over the next 4 to 6 years(Paine and Vadas 1969, Foreman 1977). Thus N. luetkeana maybe in low abundance at Bath Island due to the active grazingof the sea urchins, and absent from Sear Island due to itsreplacement by other kelp species.65105n.E 00H -5-10^I^IIIIIIIIJIIIIIIIIIIIIIIII^1967^1970^1973^1976^1979^1982^1985^1988^1991^1994YearFigure 12. Time-series of seawater surface temperatures at Active Pass, BritishColumbia, near the study sites. Temperatures are shown as the difference between themonthly mean surface temperature (*C) and the mean surface temperature over the totaltime interval (10.8 'C).Regardless of the cause, it is clear from these resultsthat there has been a shift in community structure at Bathand Sear Island since the 1970's. Although not specificallyascertained by this study, it is likely that similar changesin community structure have occurred elsewhere in the Straitof Georgia. The implications of a long-term shift inmacrophyte community structure in the Strait of Georgia areseveral fold. If an accurate assessment of environmentalimpact were needed at some location in the strait, this studyhas shown that the baseline to which the post-disturbancedata would be compared must be recent. Changes in communitystructure may have been occurring at the disturbed site inthe interim between the last sampling session and the time ofdisturbance. Comparisons to an outdated data set couldresult in an over or under-estimation of the costs of thedisturbance or in a misinterpretation the success of clean-upefforts.The possibility remains that these changes areindicative of long-term changes in environmental conditions(not necessarily restricted to temperature) in the Strait ofGeorgia, which may be of concern to parties with financial orother interest in the biota of the strait. Although thespecies observed to be changing in abundance during thisstudy are not currently of particular economic importance,other more valuable species are subject to the sameconditions, and may be experiencing changes of their own. Inthis case, the potential effects on related industries should67be anticipated, and appropriate measures taken to deal withthem. Changing conditions in the Strait of Georgia alsosuggest that more power should be granted to environmentalprotection agencies, and decisions regarding the use of thestrait should be made to err on the conservative side.One final implication of this study is the potentialfor the use of biomonitoring to indicate changing conditionsin ecological systems. Information of the sort collected inthis study could be collected in a systematic manner, withthe purpose of indicating when and where important changesmay be occurring. Time and money invested in long-termstudies aimed at monitoring and preserving the naturalresources of the Strait of Georgia would certainly paydividends by ensuring the success of industries that rely onits ecological integrity.68CONCLUSIONSThe multivariate approach of this research hasdemonstrated community level changes in benthic marinemacrophyte abundances. Given the limitations imposed on truehypothesis testing by the nature of the data set, anexploratory approach to data interpretation has been mostappropriate; it has revealed the following changes incommunity structure. 1) Along the depth gradient from 3 mabove to 6 m below zero tide level, quadrats above zero tidelevel are characterized by high abundances of Fucus spp.,Mastocarpus papillatus, Neorhodomela larix, Porphyra spp.,and Ulva spp., whereas deeper quadrats are characterized by afoliose red association (i.e. Plocamium cartilagineum,Constantinea subulifera, Prionitis spp.) and a kelpassociation (i.e. Laurencia spectablis, Corallinaofficinalis, Laminaria spp., Costaria costata). 2) Intensiveurchin grazing activity results in an overall decrease inmacrophyte variability, primarily resulting from removal ofthe foliose red and kelp species, and increases inCalliarthron tuberculosum and Ulvaria obscura. 3) Afterremoval of grazing pressure, site recovery is characterizedby an initial slight increase in Ulva spp. and Sargassummuticum during the first year, and further increases inabundance of the foliose red and kelp species in thefollowing year. 4) Long-term change in community structurehas occurred with increases in introduced species such as S.69muticum and Lomentaria hakodatensis, as well as Gigartinaexasperata, C. officinalis, and Gelidium spp., and decreasesin species such as Callophyllis spp., Cryptopleuraruprechtiana, Polyneura latissima, and Desmarestia viridis.This study has underlined the importance of long-term studiesof biotic communities in environmentally sensitive areas.Data collected during these studies, which contains recentquantitative information of the composition of local biota,is indispensable to monitor possible changes in valuablenatural resources, and to ensure that an accurate impactassessment can be carried out after a disturbance event.70BIBLIOGRAPHYBender, E. A., T. J. Case and M. E. Gilpin. 1984.Perturbation experiments in community ecology: theoryand practice. Ecology 65(1): 1-13.Chapman, A. R. 0. 1986. Population and community ecology ofseaweeds. Adv. Mar. Biol. 23: 1-161.Clayton, M. N. 1990. The adaptive significance of lifehistory characters in selected orders of marine brownmacroalgae. Aust. J. Ecol. 15: 439-452.Connell, J. H. 1987. Change and persistence in some marinecommunities. In "Colonization, succession andstability". Oxford, Blackwell Scientific Publications.Eds. A. J. Gray, M. J. Crawley and P. J. Edwards. 339-352.Crawley, M. J. 1983. Herbivory: the dynamics of animal-plant interactions. Oxford, Blackwell ScientificPublications. 437pp.Dean, T. A., S. C. Schroeter and J. D. Dixon. 1984. Effectsof grazing by two species of sea urchins(Strongylocentrotus franciscanus and Lytechinusanamesus) on recruitment and survival of two species ofkelp (Macrocystis pyrifera and Pterygophoracalifornica). Mar. Biol. 78: 301-313.Dean, T. A., K. Thies and S. L. Lagos. 1989. Survival ofjuvenile giant kelp: the effects of demographicfactors, competitors, and grazers. Ecology 70(2): 483-495.deEston, V. R. and W. O. Bussab. 1990. An experimentalanalysis of ecological dominance in a rocky subtidalmacroalgal community. J. Exp. Mar. Biol. Ecol. 136:170-195.Dethier, M. N. 1984. Disturbance and recovery in intertidalpools: maintenance of mosaic patterns. Ecol. Monogr.54(1): 99-118.DeWreede, R. E. 1983. Sargassum muticum (Fucales,Phaeophyta): regrowth and interaction with Rhodomelalarix (Ceramiales, Rhodophyta). Phycologia 22(2): 153-160.Duggins, D. O. 1980. Kelp beds and sea otters: anexperimental approach. Ecology 61(3): 447-453.71Duggins, D. 0. 1981. Sea urchins and kelp: the effects ofshort term change in urchin diet. Limn. Ocean. 26(2):391-394.Ebeling, A. W., D. R. Laur and R. J. Rowely. 1985. Severestorm disturbances and reversal of community structurein a Southern California kelp forest. Mar. Biol. 84:287-294.Elner, R. W. and R. L. Vadas. 1990. Interference inecology: the sea urchin phenomenon in the NorthwesternAtlantic. Am. Nat. 136: 108-125.Estes, J. A. and J. F. Palmisano. 1974. Sea otters: theirrole in structuring nearshore communities. Science185: 1058-1060.Estes, J. A., N. S. Smith and J. F. Palmisano. 1978. Seaotter predation and community organization in theWestern Aleutian Islands, Alaska. Ecology 59(4): 822-833.Farrell, T. M. 1991. Models and mechanisms of succession:an example from a rocky intertidal community. Ecol.Monogr. 61(1): 95-113.Foreman, R. E. 1977. Benthic community modification andrecovery following intensive grazing byStrongylocentrotus droebachiensis. Helg. wiss. Meer.30: 468-484.Foster, M. 1990. Organization of macroalgal assemblages inthe Northeast Pacific: the assumption of homogeneityand the illusion of generality. Hydrobiologia 192: 21-33.Gibson, C. W. D. and V. K. Brown. 1992. Grazing andvegetation change: deflected or modified succession?J. Appl. Ecol. 29: 120-131.Gittins, R. 1985. Canonical analysis. A review withapplications in ecology. Berlin, Springer-Verlag.351pp.Gnanadesikan, R.• 1977. Methods for statistical dataanalysis of multivariate observations. New York,Wiley. 311pp.Green, R. H. 1979. Sampling design and statistical methodsfor environmental biologists. New York, Wiley. 257pp.Harrold, C. and D. C. Reed. 1985. Food availability, seaurchin grazing, and kelp forest community structure.Ecology 66(4): 1160-1169.72Hawkes, M. W. and R. F. Scagel. 1986. The marine algae ofBritish Columbia and Northern Washington: divisionRhodophyta (red algae), class Rhodophyceae, orderRhodymeniales. Can. J. Bot. 64: 1549-1580.Hill, M. 0. 1973. Reciprocal averaging: An eigenvectormethod of ordination. J. Ecol. 61: 237-249.John, D. M., D. Lieberman and M. Lieberman. 1977. Aquantitative study of the structure and dynamics ofbenthic subtidal algal vegetation in Ghana (tropicalwest Africa). J. Ecol. 65: 497-521.Jolicoeur, P. and J. E. Mosimann. 1960. Size and shapevariation in the painted turtle. A principal componentanalysis. Growth 24: 339-354.Jongman, R. H. G., C. J. F. terBraak and O. F. R.vanTongeren, Eds. 1987. Data analysis in communityand landscape ecology. Wageningen, Pudoc. 299pp.Lindstrom, S. C. 1977. An annontated bibliography of thebenthic marine algae of Alaska. Alaska Dept. Fish andGame, Tech. Data Report 31. 172pp.Lindstrom, S. C. and R. E. Foreman. 1978. Seaweedassociations of the Flat Top Islands, British Columbia:A comparison of community methods. Syesis 11: 171-185.Littler, M. M. and S. N. Murray. 1975. Impact of sewage onthe distribution, abundance and community structure ofrocky-intertidal macro-organisms. Mar. Biol. 30: 277-291.Lubchenco, J. and B. A. Menge. 1978. Community developmentand persistence in a low rocky intertidal zone. Ecol.Monogr. 48: 67-94.LUning, K. and W. Freshwater. 1988. Temperature toleranceof northeast Pacific marine algae. J. Phycol. 24: 310-315.Paine, R. T. and R. L. Vadas. 1969. The effects of grazingby sea urchins, Strongylocentrotus spp., on benthicalgal populations. Limn. Ocean. 14: 710-719.Pearse, J. S. and A. H. Hines. 1979. Expansion of a centralCalifornia kelp forest following mass mortality of seaurchins. Mar. Biol. 51: 83-91.Pielou, E. C. 1984. The interpretation of ecological data.A primer on classification and ordination. New York,Wiley. 263pp.73Pimental, R. A. 1979. Morphometrics: The MultivariateAnalysis of Biological Data. Dubuque, Kendell/Hunt.Reed, D. C., C. D. Amster and A. W. Ebeling. 1992.Dispersal in kelps: factors affecting spore swimmingand competency. Ecology 73(5): 1577-1585.Reed, D. C., D. R. Laur and A. W. Ebeling. 1988. Variationin algal dispersal and recruitment: The importance ofepisodic events. Ecol. Monogr. 58(4): 321-335.Salmon, A. and R. H. Green. 1983. Environmentaldeterminants of unionid clam distribution in the MiddleThames River, Ontario. Can. J. Zool. 61: 832-838.Scagel, R. F. 1956. Introduction of a Japanese alga,Sargassum muticum, into the Northeast Pacific. Wash.Dep. Fish., Fish. Res. Pap. 1: 1-10.Sousa, W. P. 1980. The responses of a community todisturbance: the importance of successional age andspecies' life history. Oecologia 45: 72-81.Sousa, W. P. 1984. Intertidal Mosaics: patch size,propagule availability, and spatially variable patternsof succession. Ecology 65(6): 1918-1935.Sousa, W. P. 1985. Disturbance and patch dynamics on rockyintertidal shores. In "The Ecology of NaturalDisturbance and Patch Dynamics". Academic Press. Eds.S. T. A. Pickett and P. S. White. 101-124.Southward, A. J. 1976. On the taxonomic status anddistribution of Chthamalus stellatus (Cirripedia) inthe north-east Atlantic region: with a key to thecommon barnacles of Britain. J. Mar . Biol. Ass., UK56: 1007-1028.Tittley, I. and K. M. Shaw. 1980. Numerical and fieldmethods in the study of the marine flora of chalkcliffs. In "The Shore Environment". London, AcademicPress. Eds. J. H. Price, D. E. G. Irvine and W. F.Farnham. 213-240.Velimirov, B., J. G. Field, C. L. Griffiths and P. Zoutendyk.1977. The ecology of kelp bed communities in theBenguela upwelling system. Helg. wiss. Meer. 30: 495-518.Witman, J. D. 1985. Refuges, biological disturbance, androcky subtidal community structure in New England.Ecol. Monogr. 55(4): 421-445.74Appendix 1. Macrophyte species collected during the fiveyears of sampling.SpeciesAcrochaetium spp.Acrosiphonia saxatilis (Ruprecht) VinogradovaAgarum fimbriatum HarveyAhnfeltia fastigiata (P.&R.) MakienkoAhnfeltiopsis leptophylla (J. Agardh) Silva and DeCewAlaria tenuifolia Setchell in Collins, Holden and SetchellAmplisiphonia pacifica HollenbergAnalipus japonicus (Harvey) WynneAntithamnion defectum KylinAntithamnionella pacifica (Harvey) WollastonAntithamnionella spirographidis (Schiffer) WollastonBlidingia minima var. minima (Nageli and Kutzing) KylinBonnemaisonia nootkana (Esper) SilvaBossiella orbigniana (Descaine) SilvaBossiella californica (Descaine) SilvaBossiella chiloensis (Descaine) JohansenBossiella cretacea (Postels and Ruprecht) JohansenBossiella plumosa (Manza) SilvaBotryocladia pseudodichotoma (Farlow) KylinBranchioglossum bipinnatifidum (Montagne) WynneBryopsis plumosa (Hudson) C.A. AgardhCallophyllis flabellulata HarveyCallophyllis heanophylla SetchellC. firma (Kylin) NorrisCallithamnion spp.Calliarthron tuberculosum (P.&R.) DawsonCallophyllis violacea J. AgardhCaulocanthus ustulatus (Mertens ex Turner) KutzingCeramium californicum J. AgardhCeramium strictum HarveyCeramium pacificum (Collins) KylinChaetomorpha californica Collins in Collins,Holden and SetchellCladophora microcladioides CollinsC. seriacea (Hudson) KutzingColonial diatomsColpomenia peregrina (Sauvageau) HamelC. bullosa (Saunders) YamadaConstantinea subulifera (Setchell)Corallina frondescens Postels and RuprechtCorallina officinalis var. chilensis (Descaine in Harvey) KutzingCostaria costata (Turner) Saunders75Cryptonemia borealis KylinCryptonemia obovata J. AgardhCryptopleura ruprechtiana (J. Ag.) KylinC. lobulifera (J. Ag.) KylinCryptosiphonia woodii J. AgardhDelessericeaeDerbesia marina (Lyngbye) SolierDesmarestia ligulata (Lightfoot) LamourouxDesmarestia viridis (Muller) LamourouxDictyota binghamiae J. AgardhEctocarpus spp.Endocladia muricata (Postels and Ruprecht) J. AgardhEnteromorpha intestinalis (Linnaeus) Link in Nees von EsenbeckErythrotrichia carnea (Dillwyn) J. AgardhErythrocladia irregularis f. subingtegra (Rosenvinge) Garbary,Hansen and ScagelFarlowia mollis (Harvey et Bailey)Farlow et Setchell in CollinsFauchea spp.Fryeella gardneri (Setchell) KylinFucus spp.Fucus gardneri SilvaFucus spiralis LinnaeusGastroclonium subarticulatum (Turner) KutzingGelidium coulteri HarveyG. purpurescens GardnerGigartina exasperata Harvey and BaileyGomontia polyrhiza (Lagerheim) Bornet and FlahaultGonimophyllum skottsbergii SetchellGoniotrichopsis sublittoralis Smith in Smith and HollenbergGracilariopsis lemaneiformis (Bory) Dawson, Acleto and FoldvikGracilaria pacifica AbbottGrateloupia doryphora (Montagne) HoweGrateloupia pinnata (Postels and Ruprecht) Setchell in Collins,Holden and SetchellGrateloupia setchellii KylinGriffithsia pacifica KylinGymnogrongus chiton (Howe) Silva and DeCew in SilvaHalosaccion glandiforme (Gmelin) RuprechtHalymenia gardneri (Kylin) ParkinsonHaraldiophyllum mirable (Kylin) ZinovaHerposiphonia plumula (J. Ag.) HollenbergHeterosiphonia densuiscula KylinHildenbrandia spp.Hincksia ovata (Kjellman) SilvaHollenbergia subulata (Harvey) WollastonHYmenena spp.Iridaea heterocarpa Postels and RuprechtIridaea splendens (Setchell and Gardner) PapenfussJanczewskia gardneri Setchell and Guernsey in SetchellKallymenia oblongifructa (Setchell) Setchell76Laminaria spp.Laminaria farlowii SetchellLaminaria bongardiana Postels and RuprechtLaminaria saccharina (Linnaeus) LamourouxLaurencia spectablis Postels and RuprechtLeachiella pacifica KugrensLeathesia difformis (Linnaeus) AreschougLithothamnion spp.Lithothrix aspergillum GrayLomentaria hakodatensis YendoMastocarpus papillatus (C.A. Agardh) KylinMembranoptera tenuis KylinMicrocladia borealis RuprechtMicrocladia coulteri HarveyNeodilsea borealis (Abbott) LindstromNeorhodomela larix (Turner) MasudaNereocystis luetkeana (Mertens) Postels and RuprechtNienburgia andersoniana (J. Ag.) KylinOdonthalia floccosa (Esper) FalkenbergOpuntiella californica (Farlow) KylinPalmaria mollis (Setchell and Gardner) van der Meer and BirdPetelonia fascia (Muller) KuntzePeyssonelia pacifica KylinPhycodrys spp.Pikea californica HarveyPlatysiphonia clevelandii (Farlow) PapenfussPlatythamnion pectinatum KylinPlatythamnion reversum (Setchell and Gardner) KylinPlatythamnion villosum KylinPleonosporium vancouverianum J. AgardhPlocamium cartilagineum (L.) Dix.Polyneura latissima (Harvey) KylinPolysiphonia scopulorum var. vdllum (J. Agardh) HollenbergPolysiphonia spp.Porphyra spp.Prionitis spp.Pterosiphonia dendroidea (Montagne) FalkenbergP. bipinnata (P.&R.) FalkenbergPterosiphonia gracilis KylinPterosiphonia hamata SinovaPugetia fragilissima KylinPunctaria spp.Pilayella tenella Setchell and GardnerP. littoralis (Linnaeus) KjellmanRalfsia fungiformis (Gunner) Setchell and GardnerRhizoclonium riparium (Roth) HarveyRhodoglossum affine (Harvey) KylinRhodymenia californica KylinRhodymenia pertusa (P.&R.) J. AgardhRhodoptilum plumosum (Harvey and Bailey) Kylin77Rhodoglossum roseum (Kylin) SmithSarcodiotheca furcata (Setchell and Gardner) KylinSarcodiotheca gaudichaudii (Montagne) GabrielsonSargassum muticum (Yendo) FensholtSchizymenia pacifica KylinScytosiphon lomentaria (Lyngbye) J. AgardhSphacelaria rigidula KutzingS. racemosa GrevilleS. norrisii HollenbergStenogramma interrupta (C. Ag.) MontagneStylonema cornu-cervi ReinschSyringoderma phinneyi Henry and MullerTiffaniella snyderae (Farlow) AbbottUlva spp.Ulvaria obscura (Kutzing) Gayral var. blyttii (Areschoug)BlidingUrospora spp.Weeksia coccinea (Harvey) Lindstrom78


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items